CanoVerse: 3D Object Scalable Canonicalization and Dataset for Generation and Pose

TL;DR

CanoVerse introduces a large-scale canonical 3D object dataset and a fast canonicalization method, enabling improved pose consistency, 3D generation stability, and zero-shot orientation estimation across diverse categories.

Contribution

The paper presents a scalable canonicalization framework and a massive dataset that significantly enhance 3D object alignment, generation, and retrieval tasks.

Findings

Improved 3D generation stability

Enabled precise cross-modal shape retrieval

Achieved zero-shot point-cloud orientation estimation

Abstract

3D learning systems implicitly assume that objects occupy a coherent reference frame. Nonetheless, in practice, every asset arrives with an arbitrary global rotation, and models are left to resolve directional ambiguity on their own. This persistent misalignment suppresses pose-consistent generation, and blocks the emergence of stable directional semantics. To address this issue, we construct \methodName{}, a massive canonical 3D dataset of 320K objects over 1,156 categories -- an order-of-magnitude increase over prior work. At this scale, directional semantics become statistically learnable: Canoverse improves 3D generation stability, enables precise cross-modal 3D shape retrieval, and unlocks zero-shot point-cloud orientation estimation even for out-of-distribution data. This is achieved by a new canonicalization framework that reduces alignment from minutes to seconds per object via compact hypothesis generation and lightweight human discrimination, transforming canonicalization from manual curation into a high-throughput data generation pipeline. The Canoverse dataset will be publicly released upon acceptance. Project page: https://github.com/123321456-gif/Canoverse